Kokshoorn, N.E.
Citation
Kokshoorn, N. E. (2011, December 7). Hypopituitarism : clinical assessment in different conditions. Retrieved from https://hdl.handle.net/1887/18194
Version: Corrected Publisher’s Version
License: Licence agreement concerning inclusion of doctoral thesis in the Institutional Repository of the University of Leiden
Downloaded from: https://hdl.handle.net/1887/18194
Note: To cite this publication please use the final published version (if applicable).
Chapter 2
Clinical review: Hypopituitarism following traumatic brain injury – the prevalence is aff ected by the use of diff erent dynamic tests and diff erent normal values
Nieke E. Kokshoorn, Moniek J.E. Wassenaar, Nienke R. Biermasz, Ferdinand Roelfsema, Johannes W.A. Smit, Johannes A. Romijn and Alberto M. Pereira
Eur J Endocrinol. 2010 Jan;162(1): 11–18
38
Abstract
Objective: Traumatic brain injury (TBI) has emerged as an important cause of hypopituitarism. However, considerable variations in the prevalence of hypopituitarism are reported. Th ese can partly be explained by severity of trauma and timing of hormonal evaluation, but may also be dependent on endocrine tests and criteria used for diagnosis of hypopituitarism.
Methods: Systematic review of studies reporting prevalence of hypopituitarism in adults ≥ 1 year aft er TBI focusing on used (dynamic) tests and biochemical criteria.
Results: We included data from 14 studies with a total of 931 patients.
Th ere was considerable variation in defi nition of hypopituitarism.
Overall, reported prevalences of severe GH defi ciency varied between 2 and 39%. Prevalences were 8–20% using the GHRH-arginine test (cut- off < 9 μg/L), 11–39% using the glucagon test (cut-off 1–5 μg/L), 2% using the GHRH test (no cut-off ) and 15–18% using the insulin tolerance test (ITT) (cut-off < 3 μg/L).
Overall, the reported prevalence of secondary adrenal insuffi ciency had a broad range from 0 to 60%. Th is prevalence was 0–60% with basal cortisol (cut-off < 220 or < 440 nmol/L), 7–19% using the ACTH test and 5% with the ITT as fi rst test (cut-off < 500 or < 550 nmol/L).
Secondary hypothyroidism was present in 0–19% (free T 4 ) or 5–15%
(TRH stimulation). Secondary hypogonadism was present in 0–29%.
Conclusion: Th e reported variations in the prevalence rates of hypopituitarism aft er TBI are in part caused by diff erences in defi nitions, endocrine assessments of hypopituitarism and confounding factors.
Th ese methodological issues prohibit simple generalizations of results of
original studies on TBI-associated hypopituitarism in the perspective of
meta-analyses or reviews.
39
Introduction
In recent years, an increasing number of studies have reported the presence of pituitary insuffi ciency in patients who experienced traumatic brain injury (TBI) (1–14). Th e prevalence of pituitary insuffi ciency aft er TBI appeared to be unexpectedly high (15;16). Remarkably, the prevalence rates varied considerably among the diff erent studies, ranging from 15 to even 90% of the patients.
Several factors infl uence the prevalence of hypopituitarism aft er TBI. First, the time interval between TBI and the assessment of pituitary function, since hormone alterations mimicking pituitary insuffi ciency are prevalent in the early post-traumatic period. Second, the type and severity of the brain injury aff ects the prevalence of hypopituitarism, because persistent pituitary insuffi ciency is only frequent aft er severe TBI (7;15). Th ird, endocrine tests, assays, and criteria for the diagnosis of hypopituitarism diff er between the studies. Although many reviews have addressed TBI-related hypopituitarism, a detailed comparison of these methodological issues between the diff erent studies has not been performed for each pituitary axis.
We hypothesized that these methodological diff erences may have
contributed, at least in part, to the discrepancies in prevalence rates of
hypopituitarism aft er TBI, reported by the diff erent studies. Th erefore,
the aim of this study was to critically compare the pituitary function
tests, and defi nitions of hypopituitarism between studies that assessed
the long term outcome of TBI on pituitary function.
40
Patients and methods
Search strategy
We performed a search in PubMed, EMBASE, Web of Science, and the Cochrane database, for all published studies on the association between TBI and hypopituitarism. Th e following search strategy was used:
(Traumatic Brain Injury OR Traumatic brain injuries) AND (traumatic OR trauma) AND (Hypopituitarism OR Hypopituitar* OR Hypothalamus Hypophysis System OR “Hypothalamopituitary dysfunction” OR “pituitary dysfunction” OR Hypothalamo-Hypophyseal System OR Pituitary Gland OR Hypophysis).
In addition, the references of relevant articles were checked for additional articles. Th e search was performed on 23 March 2009. Only original articles were included. We used the following exclusion criteria:
pediatric or adolescent population, publications concerning pituitary testing < 12 months aft er injury (a median of 12 months was accepted), and articles that evaluated pituitary insuffi ciency aft er subarachnoidal bleeding (SAB).
Data review
Th e following data were extracted from each study: 1) age and gender, 2) the endocrine tests used for assessment of each pituitary axis, 3) defi nitions used for pituitary insuffi ciency for each pituitary axis, 4) hormone assays, 5) reference values provided in the manuscript, and 6) use of control populations. Tables were constructed per pituitary axis.
Th ese tables are added as supplemental data fi les. Th e growth hormone
(GH)-IGF-I axis (Table 2), the pituitary-adrenal axis (Table 3), the
pituitary-thyroidal axis (Table 4), the pituitary-gonadal axis (Table 5),
and prolactin (Table 6).
41
Results
We identifi ed 278 articles, of which 218 were excluded on the basis of
title and abstract. Of the remaining 60 articles, 46 were reviews. Finally,
14 original studies were included with a total of 931 patients. Details
of these studies are summarized in Table 1. Th e number of patients
evaluated by the diff erent studies varied between 22 and 105.
42
T a b le 1 . S tu d ie s o n T B I a n d p it u it a ry d e fi ci e n c y S tudy Y ear of public a tion N o . of pa tien ts T ime of t esting p ost TBI [mon ths (median)]
T rauma se v erit y (GCS) BMI (k g/m
2)
A n y pituitar y defi cienc y (%) Kelly et al .(6) 2000 22 3–276 (median 26) 3–15 25.1±6.5 37 Lieber man et al .(9) 2001 70 1–276 (median 13) 3–15 84% GCS ≤ 8 NR 69 B ondanelli et al .(3) 2004 50 12–64 3–15 54% GCS ≤ 8 24.6±0.4 54 A gha et al .(1) 2004 102 6–36 (median 17) 3–13 56% GCS ≤ 8 NR 28 P opo vic et al .(10) 2004 67 12–264 3–13 24.8±0.5 34 A imar etti et al .(2) 2005 70 12 3–15 21% GCS ≤ 8 23.8±0.4 23 Leal- C er ro et al .(8) 2005 99 >12 ≤ 8 25.2±3.0 (n=44) 25 S chneider et al .(11) 2006 70 12 3–15 23.8±3.2 36 Tanr iv er di et al .(12) 2006 52 12 3–15 25% GCS ≤ 8 NR 51 Her rmann et al .(5) 2006 76 5–47 ≤ 8 25.8±4.2 24 Bushnik et al .(4) 2007 64 > 12 mon ths NR NR 90 K lose et al .(7) 2007 104 10–27 (median 13) 3–15 38% GCS ≤ 8 25* (17–39) 15 Tanr iv er di et al .(14) 2008 30 36 3–15 16.7% GCS ≤ 8 NR 30 W ach ter et al .(13) 2009 55 NR 3–15 17% GCS ≤ 8 NR 25 Total No . of pa tien ts: 931 B M I, b o d y m a ss in d e x re p o rt e d a s th e m e a n ± S E M ; G C S , G la sg o w C o m a S ca le so re ; T B I, tr a u m a ti c b ra in in ju ry ; N R , n o t re p o rt e d . * re p o rt e d a s m e d ia n (r a n g e )
43 The GH-IGF-I axis
Th e prevalence of GH defi ciency (GHD) ranged between 2 and 66%
(severe GHD 39%; Figures 1 and 2 and Suppl Table 2). Th e presence of GHD was associated with higher body mass index (BMI) values in some of the studies (Figure 1). In addition to basal serum GH and IGF-I values, all studies used a dynamic test to assess GH secretory reserve. However, diff erent dynamic tests were used.
Th ree studies (196/931=21% of all patients) used the combined GHRH- arginine test as the fi rst screening. Th e criterion for severe GHD was a peak GH level < 9.0 μg/L in all three, which was not adjusted for BMI.
Prevalence rates of severe GHD varied between 8 and 20% (weighted mean 12%) (2;3;5). Schneider et al. (11) also used the GHRH-arginine test, but only in a subset of the patients (those with abnormal serum cortisol levels, n=32); the prevalence of GHD in this study was 10%.
Two studies (112/931=12% of all patients) used an insulin tolerance test (ITT) as the primary screening test (6;7). Th e criterion for severe GHD was a peak GH response < 3 μg/L in both, and the prevalence of GHD was comparable (18 and 15% respectively; weighted mean 16%).
Of the eight remaining studies, three used a stimulation test with glucagon (n=209) (1;4;9) with prevalence rates for severe GHD between 11 and 39% (weighted mean 20%). Th e cut-off values diff ered considerably and varied between 1 and 5 µg/L between these studies. Just one study used a stimulation test with GHRH only (number of patients not recorded) reporting a GHD prevalence of 2% (13). Two studies (n=119) used the combined GHRH-GHRP6 test with a prevalence of 15 and 33%
respectively (weighted mean 21%) (10;12). Th e cut-off values were similar (GH < 10 μg/L) within these studies, and were derived from another report (17).
Finally, two studies used a combination of these tests (8;14). For
instance, Agha et al. (1) used a glucagon stimulation test for the initial
screening in 102 subjects, and in case of incomplete GH response, they
used an ITT (n=14) or combined GHRH plus arginine test (n=4) to
confi rm GHD.
44
= GHD = N o GHD 3 2
3 6 * *
* 2 4
2 8
n=7 n=10
* 2 0
n=41 n=91n=70 n=88
n=23
n=16 n=420
Body mas s in de x ( BM I) (k
2
g/m
) 0 W eigh ted a h to ta l mea n (± 9 5 % C I)
n=47
n=17 n=6
n=7n=84
b c d e f g
#n=34
n=11 n=10 n=26
F ig u re 1 . B o d y m a ss in d e x in p a ti e n ts d ia g n o se d w it h v e rs u s th o se w it h o u t G H d e fi ci e n c y (d a ta a v a il a b le o n ly in 8 o u t o f th e 1 6 st u d ie s) . a , L ie b e rm a n et a l. 9 ); b , A g h a et a l.( 1) ; c , P o p o v ic et a l. (1 0 ); d , L e a l- C e rr o et a l. (8 ); e , T a n ri v e rd i e t a l.( 12 ); f, H e rr m a n n et a l.( 5 ); g , K lo se et a l.( 7 ); h , T a n ri v e rd i e t a l.( 1 4 ); i, w e ig h te d to ta l m e a n (m e a n ± 9 5 % C I: n o G H D 2 4 .5 (2 4 .2 –2 4 .9 ) v e rs u s G H D 2 7. 7 (2 6 .7 –2 8 .8 ) k g /m
2). G H D , G H d e fi ci e n c y ; B M I, b o d y m a ss in d e x ; * P < 0 .0 5 .
#D a ta re p o rt e d a s m e d ia n w it h r a n g e ; n o t i n cl u d e d i n t h e t o ta l w e ig h te d m e a n .
45
GHRH-arginine as initial test
15 20 25
GHRH-arginine as confirmation test n=70
A
15 20
25 ITT as initial test
Weighted
*
**
B
Weighted mean 13%
0 5 10 15
Bondanelli Aimaretti Herrmann Schneider
n=50 n=76
n=32
0 5 10 15
Kelly Klose n=22
n=104
mean 16%
P re v a le n ce s e v e re G H D ( % ) P re v a le n ce s e v e re G H D ( % )
25 30 35
n=32
GHRH-GHRP6 as initial test C
35 40 45
n=59
Glucagon stimulation test as initial test
**
D
5 10 15 20 25
n=67
Weighted mean 21%
5 10 15 20 25 30
n=102
n=70
Weighted mean 21%
* *
P re v a le n ce s e v e re G H D (%) P re v a le n ce s e v e re G H D (%)
0
Popovic et al.
Tanriverdi et al.
0 Agha
et al.
Bushnik et al.
Lieberman et al.
E Combination of tests
25 Δ Δ
5 10 15 20
GHRH test
n=99
n=30
Δ
Pr e v al e n ce s e v e re G H D ( % )
0 Leal-Cerro et al.
Tanriverdi et al.
Wachter et al.
P n=55
et al. et al. et al. et al. et al. et al.
Figure 2. Absolute and weighted mean prevalence rates of severe GH defi ciency (GHD) according to the stimulation tests used per study. The number of patients tested is depicted in each bar. Panel A: the combined GHRH-arginine test; defi nition severe GHD:
peak GH < 9 μg/L for all four studies. Panel B: the insulin tolerance test (ITT); *defi nition severe GHD: GH < 95%
CL according to AUC; **defi nition severe GHD: peak GH < 3 μg/L. Panel C: the combined GHRH-GHRP6 test; defi nition severe GHD: peak GH < 10 μg/L for both studies. Panel D: the glucagon stimulation test;
defi nition severe GHD: *peak GH < 3 μg/L; **peak GH
< 5 μg/L. Panel E: combined stimulation tests as initial
screening followed by confi rmation test; ΔGHRH-
GHRP6 test as initial test; ITT and glucagon stimulation
test as confi rmation tests; ΔΔGHRH-GHRP6 test as initial
test; glucagon stimulation test as confi rmation test.
46
The Pituitary-Adrenal axis
Th e prevalence of secondary adrenal insuffi ciency defi ciency ranged from 0 to 60% between the studies (Figure 3, Suppl Table 3).
Four studies (251/931=27% of all patients) only measured basal morning fasting serum cortisol and/or ACTH levels (2–4;10), resulting in prevalence rates between 0 and 60% (weighted mean 15%). Th e criteria for pituitary-adrenal insuffi ciency diff ered between three studies (cortisol < 220–440 nmol/L), and were not reported in the fourth study (10). Th e study reporting the highest prevalence of 60% used a cut-off value of 440 nmol/L (4).
Four studies (145/931=12% of all patients) used an ACTH stimulation test (Synacthen; either with 1 or 250 μg). However, only one study performed this test in all patients and the prevalence of ACTH defi ciency was 7% (9). In the other three studies, only a subset of the patients (those with subnormal basal cortisol levels) underwent stimulation with ACTH.
Th e prevalence in these studies varied between 7 and 19% (weighted mean 10%) (11;12;14).
One study (55/931=6% of all patients) used nonstimulated cortisol values between 1600 and 2000 h (reference values 63–339 nmol/L), which was followed by a corticotrope releasing hormone (CRH) test only in those with values below this reference range, or in those who responded confi rmatory to a specifi c questionnaire (13).
In the remaining fi ve studies (403/931=43% of all patients), the ITT
was used in 169 patients as a primary test (n=112) resulting in a prevalence
of 5% in both studies (6;7), or as a confi rmation test in a subset of the
patients. Two studies measured basal serum cortisol levels and used ITT
as a confi rmation test (prevalence of 3 and 11% respectively) (5;8). One
study assessed primarily with a glucagon stimulation test (n=102), and
used the ITT and ACTH tests to confi rm ACTH defi ciency (prevalence
13%) (1). Th e criteria for a normal cortisol response to hypoglycemia
were a peak cortisol level > 550 nmol/L in one (8), and > 500 nmol/L
in three other studies (1;5;7). Th e fi ft h study used a control group of
18 healthy subjects to defi ne normal cortisol responses to ITT (cortisol
response < 95% confi dence limit according to the obtained area under
the curve) (6). Th e CRH test was used in only one study and did not
report the number of patients (13).
47
50 60 70
Basal serum ACTH and cortisol levels
n=64
***
A
16 18 20
ACTH test in total population
n=52
ACTH test in subset of population
**
B
10 20 30 40 50
P re v a le n ce A C T H d e fi ci e n cy ( % )
Weighted mean 15%
* ** ** 4
6 8 10 12 14
P re v a le n ce A C T H d e fi ci e nc y ( % )
Weighted mean 10%
n=70
n=70
n=30
*
*
**
Bondanelli et al.
0
Popovic et al.
Aimaretti et al.
Bushnik et al.
n=50 n=67 n=70
0 2
Lieberman et al.
Schneider et al.
Tanriverdi et al.
Tanriverdi et al.
C
12 ITT as initial test ITT as confirmation test
14 Other tests D
6 8 10 12
e ACTH defi ci e nc y (%)
Weighted
* **
***
n=99
6 8 10 12 14
e A CT H d e fi ci e n cy ( % )
n=102